Problem Set #6
Math 471 Real Analysis
Assignment: Chapter 2 #4, 16
Chapter 3 #3, 21
Chapter 4 #4, 8, 12
Clayton J. Lungstrum
October 11, 2012
Exercise 2.4
Let {fk } k=1 be a sequence of functions of bounded variation on [a, b]. If V [fk ; a, b] M <
for all k and if fk f pointwise on [a, b], show that f is of bounded variation and that
V [f ; a, b] M . Give an example of a convergence sequence of functions of bounded variation
whose limit is not of bounded variation.
Solution.
Let > 0 and let = {xi }N
i=0 be a partition of [a, b]. Then, from the hypotheses, we have
N
X
|fk (xi ) fk (xi1 )| M
i=1
for all k N. Now, since fk f pointwise, there exists a k0 such that |f (x0 )fk0 (x0 )| < 2N .
Likewise, we can do this for each of x1 , . . . , xN , since there are only finitely many. Then let
k = max{k0 , k1 , . . . , kN }. Thus, we have
N
X N
X
|f (xi ) f (xi1 )| = |f (xi ) fn (xi ) + fn (xi ) fn (xi1 + fn (xi1 ) f (xi1 )|
i=1 i=1
XN N
X
|f (xi ) fn (xi )| + |fn (xi ) fn (xi1 )|
i=1 i=1
N
X
+ |f (xi1 ) fn (xi1 )|
i=1
N N
X X
< +M +
i=1
2N i=1
2N
= +M +
2 2
= M + ,
where n > k. Since > 0 was arbitrary, we have M as an upper bound for N
P
i=1 |f (xi )
f (xi1 )|, thus we may take the supremum over all partitions and we have that M is an upper
bound (since the supremum is the least upper bound), hence f is of bounded variation and
has total variation at most M . To see that we cannot relax the condition where the fk s are
of uniformly bounded variation (i.e., the upper bound does not depend on k), consider the
following of sequence of functions which is uniformly convergent,
(
x sin x1
2
: x n ,1
fn (x) = 2 .
0 : x 0, n
Clearly fn f uniformly, where f is the function defined in Exercise 1, which is of un-
bounded variation. Thus, even tightening the convergence hypothesis does not grant that
the limit function be of bounded variation if the sequence of functions is not of uniformly
bounded variation.
Q.E.D.
1
Exercise 2.16
Suppose that is of bounded variation on [a, b] and that f is bounded and continuous
except for a finite number of jump discontinuities in [a, b]. If is continuous at each discon-
Rb
tinuity of f , show that a f d exists.
Solution.
Since is of bounded variation, without loss of generality, we can assume it is increasing
(Jordans Theorem). Let > 0 and let M = supx[a,b] |f (x)|. We know M < since f
is bounded. Also, let E be the finite set of discontinuities of f . Then we can cover E by
finitely many disjoint intervals [uj , vj ] placing the intervals so that for each discontinuity of
f in (a, b), the point of discontinuity is in the interior of one the [uj , vj ] intervals. Also,
note that since is continuous at the discontinuities of f , we can place the intervals so that
(vj ) (uj ) < . Now, if we remove the segments (uj , vj ) from [a, b], we have a compact
subset, hence f is uniformly
SN continuous; that is, there is a > 0 such that |f (x) f (y)| <
for all x, y [a, b] j=1 (uj , vj ) such that |x y| < .
Now form a partition in the following way: let = {xn }Q n=1 where each uj and vj are
in the partition and no point from any of the segments (uj , vj ) occur, and xi < unless
xn1 = uj . Then, since Mn mn < 2M for every n and Mn mn unless xn1 = uj
(where Mn = supx[xn1 ,xn ] f (x) and mn defined analogously), we have
U (, f, ) L(, f, ) ((b) (a)) + 2M .
Since > 0 was arbitrary, we have the f is Riemann-Stieltjes integrable with respect to .
Q.E.D.
2
Exercise 3.6
Construct a Cantor-type subset of [0, 1] by removing from each interval remaining at the
k th stage a subinterval
Pof relative length k , 0 < k < 1. Show that the remainder has measure
zero if and only if k=1 k = .
Solution.
Let 0 = T
[0, 1]. Then let k be the set remaining after k iterations of the above process
and let = k=1 k . Clearly is measurable as the countable intersection of measurable
sets. Also
Qn note that , like the Cantor set, consists of 2n disjoint closed intervals, each of
1k
length k=1 2k , so then we have
n
Y
|n | = (1 k ).
k=1
Now, each factor is strictly less than one, hence |k | is a strictly decreasing sequence bounded
below by zero, thus converges to some real number, and by the finiteness of each |k |, we
have limkP |k | = ||. Now, either limk k 0 or limk k c > 0. In the latter
case, clearly k=1 k = and we can calculate
n
Y n
Y
|n | = (1 k ) (1 c) = (1 c)n
k=1 k=1
which goes to 0 as n tends to .
In the former case, we must apply lHospitals rule in the following way:
k x 1
lim = lim+ = lim+ 1 = 1.
k log(1 k ) x0 log(1 x) x0
1x
Thus, since the limit exists and is finite, taking the absolutePvalue, the limit comparison test
implies
P that both either converge or diverge; in particular, k=1 k converges if and only if
Pk=1 log(1 k ) converges. Thus, by the final hint, limn |k | is nonzero if and only if
k=1 log(1 k ) converges.
Q.E.D.
3
Exercise 3.21
Show that there exist sets E1 , E2 , . . . , Ek , . . . such that E =
T
k=1 Ek , |Ek |e < , and
limk |Ek |e > |E|e with strict inequality.
Solution.
Let An =
S
k=n A + qk , where A is a nonmeasurable subset of [0, 1] and {qk }k=1 is an
enumeration of the rational number in [0, 1]. Then, as before,
2 lim |An | |A| > lim An = 0.
n n
Thus, we simply apply the previous exercise (from the text) in a reverse order.
Q.E.D.
4
Exercise 4.4
Let f be defined and measurable in Rn . If T is a nonsingular linear transformation of
n
R , show that f (T x) is measurable.
Solution.
Let
E1 = {x : f (x) > a} and E2 = {x : f (T x) >}.
Observe that since T is linear and nonsingular, we have the following:
T 1 E1 = {T 1 x : f (x) > a}
= {T 1 x : f (T (T 1 x)) > a}
= {y : f (T y) > a}
= E2 ,
thus, f (T x) is measurable as the inverse image of a measurable set is measurable under
linear transformations.
Q.E.D.
5
Exercise 4.8
(a) Let f and g be two functions which are usc at x0 . Show that f + g is usc at x0 . Is
f g usc at x0 ? When is f g usc at x0 ?
(b) If {fk }
k=1 is a sequence of functions which are usc at x0 , show that inf kN fk (x0 ) is usc
at x0 .
(c) If {fk }
k=1 is a sequence of functions which are usc at x0 and which converge uniformly
near x0 , show that limk fk is usc at x0 .
Solution.
(a) Let us assume that neither f nor g is , for the result would then be trivial. Then
lim sup(f (x) + g(x)) lim sup f (x) + lim sup g(x) f (x0 ) + g(x0 ) = (f + g)(x0 ),
xx0 xx0 xx0
and so f + g is usc if both f and g are usc. However, f g is not necessarily usc if
both f and g are usc. Consider the following:
(
0 : x R \ {0}
f (x) = 0 : x R and g(x) = .
1 : x=1
Then f + g = g is clearly usc as we saw in class, while f g = g is not usc, also as we
saw in class. Note that using the same principle, we can see sup{(f g)(x) : x x0 }
sup{f (x) : x x0 } sup{g(x) : x x0 } only if f and g are nonnegative functions,
i.e.,
lim sup(f g)(x) lim sup f (x) lim sup g(x) f (x0 )g(x0 ) = (f g)(x0 ),
xx0 xx0 xx0
hence, f g is usc.
(b) Let f (x) = inf kN fk (x) and observe
lim sup f (x) lim sup fk (x) fk (x0 )
xx0 xx0
for all k N. Thus,
lim sup f (x) inf fk (x0 ) = f (x0 ),
xx0 kN
therefore, f is usc.
(c) Let f (x) = limk fk (x). Then, for > 0, there exists a k N such that supxE {|f (x)
fk (x)|} < . Then
lim sup f (x) < lim sup fk (x) + fk (x0 ) + < f (x0 ) + 2
xx0 xx0
for arbitrary > 0, hence f (x) is usc at x0 .
Q.E.D.
6
Exercise 4.12
If f (x), x R is continuous at almost every point of an interval [a, b], show that f is
measurable on [a, b]. Generalize this to functions defined in Rn .
Solution.
NOTE: This proof does not rely on the fact that [a, b] is a closed, bounded interval.
Let f be continuous a.e. on [a, b], then there is a set of measure zero, say N , such that
f is continuous on [a, b] N . Since N is measurable, for every natural number T n, there is
1 n
a corresponding open set Gn such that N GT n and |Gn | < n . Writing Gn = k=1 Gk if
necessary, we can assume Gn+1S Gn . Let Q = G
n=1 n
; then N Q and |Q| = 0. Now, f
n
is continuous on Fn = [a, b] k=1 Gk , so if I is any open interval,
f |1
Fn (I)
is a relatively open set in Fn , hence is measurable. Since (
S
n=1 Fn ) Q is the whole space,
we have for any open interval I,
!
[ 1 1
f 1 (I) = f (I) f (I),
Fn Q
n=1
which is a countable union of measurable sets, thus is measurable, i.e., f is measurable.
Q.E.D.
